Pup Mortality in Laboratory Mice

(1)

Pup Mortality in Laboratory Mice

Influence of Maternal Behaviour and Housing Environment

Elin M. Weber

Faculty of Veterinary Medicine and Animal Science Department of Animal Environment and Health

Skara

Doctoral Thesis

Swedish University of Agricultural Sciences

Skara 2015

(2)

Acta Universitatis agriculturae Sueciae

2015:6

ISSN 1652-6880

ISBN (print version) 978-91-576-8210-9

Cover: Mouse pup struggling to get back into the nest (Photo: Elin Weber)

(3)

Pup Mortality in Laboratory Mice. Influence of Maternal Behaviour and Housing Environment

Abstract

Successful mouse breeding is a crucial part of providing animals for research.

However, loss of single pups or entire litters after birth is a relatively common problem.

Determining how pups die is crucial for the understanding of mortality, but the scientific literature does not provide a clear picture of pup mortality and the reason why pups die is still poorly understood.

The overall aim with this thesis was to investigate the causes of pup mortality in laboratory mice, focusing on maternal behaviour and the effect of housing environment. Specifically the aims were to investigate if litter loss was higher in primiparous females (study 1), if female mice actively killed their pups and if there were any differences in behaviour between females that lost the litter shortly after birth and females that successfully weaned their litters (study 2), and how the conditions for nest building influenced nest building and pup survival (study 3).

In study 1 (paper I), breeding data from mice of the strains C57BL/6 and BALB/c were used. An effect of strain but no effect of parity on litter mortality was found. In study 2, C57BL/6 females were housed in four different treatments with different amounts of nesting material and cage furnishment. Behaviours of females whose litter died were observed in detail from birth of the litter until the litter died (paper II). No evidence that females actively killed their pups was found. In paper III, both females that lost their entire litter shortly after birth and females that successfully weaned their litter were observed from 24h before to 24h after parturition. Litter loss was associated with females showing less nest-building behaviour before parturition, more parturition- related behaviours and more time outside the nest. In the last study (paper IV) females were housed in four treatments with different amount of nesting material and structure present or absent. Females given a larger amount of nesting material built more dome shaped nests of higher quality.

In summary, this thesis does not support the assumption that female mice actively kill their offspring. Pregnant females should be given a large amount of nesting material to enable nest-building behaviour. Further, monitoring females around time for parturition should be considered to detect problematic parturitions.

Keywords: maternal behaviour, postnatal mortality, nest building, nesting material, laboratory mice, breeding, pup survival

Author’s address: Elin Weber, SLU, Department of Animal Environment and Health, P.O. Box 324, SE-532 23 Skara, Sweden

E-mail: Elin.Weber@ slu.se

(4)

Dedication

To Robert and Piri

and to all the mice out there…

The first step toward change is awareness.

Nathaniel Branden

(5)

List of Publications

This thesis is based on the work contained in the following papers, referred to by Roman numerals in the text:

I Weber, E.M., Algers, B., Würbel, H., Hultgren, J., Olsson, I.A.S. (2013).

Influence of Strain and Parity on the Risk of Litter Loss in Laboratory Mice. Reproduction in Domestic Animals 48, 292-296.

II Weber, E.M., Algers, B., Hultgren, J., Olsson, I.A.S. (2013). Pup mortality in laboratory mice – infanticide or not? Acta Veterinaria Scandinavica 55, 1-8.

III Weber, E.M., Hultgren, J., Algers, B., Olsson, I.A.S. Mortality in laboratory mouse pups – do females that lose their litters behave differently? (Manuscript)

IV Weber, E.M., Hultgren, J., Olsson, I. A. S., Algers, B. Nest quality and pup survival in laboratory mice given different conditions for maternal nest building. (Manuscript)

Papers I-II are reproduced with the permission of the publishers.

(8)

(9)

1 Introduction

Mice are incredible animals. With their amazing ability to adapt and eat a wide variety of food, they inhabit almost every ecological niche in the world and are the most widespread mammal on the planet. The house mouse (Mus musculus) has lived in close association with human populations since civilization first appeared (Silver, 1995). Selectively bred mice are also highly valued as experimental animals and have been used in research for nearly 100 years (Harper, 2010), today constituting the most commonly used vertebrate species in biomedical research. Seven million mice were reported being used in the European Union 2011 (Seventh Report from the Commission to the Council COM(2013) 859 final) and approximately 25 million mice used worldwide (Harper, 2010).

Their short reproductive cycle, short life span, small size and low maintenance costs are characteristics that have contributed to the mice becoming the most popular mammal in research (Baumans, 2004). A crucial part of providing animals for research is successful breeding. Still, in many facilities breeding efficiency is complicated by problems with reproduction, such as pre-weaning pup mortality. Both single pups and the entire litter can be lost, with loss of entire litters having the most substantial influence on breeding efficiency. It is sometimes assumed that it is normal for laboratory mice to lose their first litter due to the mother being inexperienced. Since dead mouse pups are often eaten by their mother, there is further a widespread belief that the female actively kills them. However, the scientific support for these assumptions and the overall understanding of what causes pup loss and how mouse pups die is still insufficient. To increase this understanding, and potentially improve welfare of laboratory mice, it is important to consider the behavioural biology of the wild house mouse (Latham & Mason, 2004), and how their needs might be influenced by the housing conditions provided in the laboratory.

(10)

1.1 Free-living house mice

Free-living mice are burrowing animals that are mainly nocturnal (Walker &

Nowak, 1999; Sayler & Salmon, 1971). They are very active and agile animals with excellent sense of balance and can run fast up almost any vertical surface or horizontally along small ropes and move hanging upside-down from 6mm hardware mesh with ease. They are excellent jumpers and can squeeze through very small openings (Roll, 2009). Mice are also highly explorative and in the wild they spend a substantial time on seeking a wide variety of food.

Mice form complex social structures (Baumans, 2004) and two common types of populations have been described, commensal and feral (Bronson, 1979). Commensal mice rely on humans for food and shelter and live in territories with stable and plentiful food supply with a population density of up to 10 mice per m². Feral populations are less dense (up to 1 mouse/m²), do not depend on humans and are found in environments with seasonally unstable food supply (Bronson, 1979).

Further, all mice build nests in which they sleep, seek shelter and take care of their offspring. Nests are built in underground burrows or hidden places above ground and are lined with grass, dried plants or other soft materials (Van Oortmerssen, 1971). When giving birth, pairs or groups of females residing within one male’s territory usually form a communal nest and also nurse their pups communally (Manning et al., 1995; Packer et al., 1992; Wilkinson &

Baker, 1988). The relatedness of the females and their offspring is probably an important aspect of communal nesting; nest mates who grow up together typically have the same father and females sharing the same nest are often related. Whether her own or the other females’ offspring, it is therefore highly likely that any pup a female nurses will be closely related offspring when communally nesting with a familiar female (König, 1994).

1.2 Housing and management in the laboratory

In the laboratory, mice are typically housed in small plastic transparent cages with wire tops, provided with bedding material and sometimes nesting material. Food from a food hopper in the wire top and water is generally provided ad libitum. Caging systems can be either open or individually ventilated and rooms are maintained with controlled dark: light cycle, temperature and humidity. Overall, the housing and husbandry practices in the laboratory have been designed to provide a standardised environment, with

(11)

preventing animals from performing many motivated behaviours and giving them little control over their environment (Olsson & Dahlborn, 2002).

However, despite being bred for hundreds of generations in the laboratory environment, mice still have a strong motivation to perform many of the behaviours seen in their wild ancestors. It has been argued that behaviours essential for survival in the wild will remain highly motivated also in animals in captive environments (Dawkins, 1998; Dawkins, 1990). Nest building is one such behavioural need; both breeding and non-breeding laboratory mice still have a strong motivation to build nests (Olsson & Dahlborn, 2002; Estep et al., 1975). They will work for access to nesting material (Roper, 1976) and when offered a choice, they show a strong preference for access to nesting material (Van de Weerd et al., 1998). When presented with different types of nesting material, mice given more “naturalistic” material built nests of higher quality (Hess et al., 2008). Access to nesting material thus enables mice to perform nest-building behaviour. It also provides shelter and thus a possibility to escape from potential stressors. To a certain degree, nesting material gives mice a chance to control the microclimate in the cage (Gaskill et al., 2011).

The laboratory cage obviously differs from the habitat of the wild mouse ancestors in several aspects also relevant to reproduction. Breeding systems used in the laboratory consists of breeding pairs (one male and one female), trios (one male and two females) or harem groups (one male and several females). In pairs and trios, the male and females are usually kept together to enable postpartum mating. In harem groups, the females are often placed in separate cages when pregnancy is confirmed. If females are housed in groups or individually when giving birth usually depends on the importance of determining which female the pups belong to. In the wild on the other hand, it would probably be rare for a female to raise a litter alone without the presence of the male or other females. Also the weaning process differs. In the laboratory, mouse pups are generally weaned at the age of 21 days, corresponding to the time when the next litter will be born if the female was mated postpartum. However, Bechard and Mason (2010) report that laboratory mouse independence occurs weeks after this age, and this early weaning age might deprive mouse pups of maternal care.

Confined in a laboratory cage, mice have very limited possibilities to adjust their environment. The amount and material of the bedding provided usually limits them to dig more than a few centimetres and thus do not enable creation of burrows. Further, they cannot choose nest site and are restricted to the nesting material provided by the laboratory staff. The nest will thus not be situated in a burrow or a hidden place but instead in a brightly illuminated room. Wallace (1981) found the provision of extensive nesting and burrowing

(12)

opportunities to be crucial for successful breeding of wild mice in the laboratory. Laboratory mice are typically also kept at a temperature of 20-24˚C which is below their thermoneutral zone (Gaskill et al., 2009). Depending on their ability to create a nest of sufficient quality, they might thus also be exposed to cold stress.

Giving animals the opportunity to perform motivated behaviours, access preferred environments and give them control over their environment are important aspects of welfare of captive animals, and are also required in the European Union according to Directive 2010/63/EU of the European Parliament and of the Council of 22 September 2010 on the protection of animals used for scientific purposes (OJ L 276, 20.10.2010, p.33).

1.3 Reproduction

The reproductive strategy of mice is to produce many large litters, and under favourable conditions, female house mice can give birth to 6-10 young every month. Reproductive performance varies widely between strains (Silver, 1995) but generally, they sexually mature around the age of 6–8 weeks with oestrus cycles that last between 4 and 6 days (Berry, 1970; Bronson et al., 1966). The length of individual cycles varies, and is influenced by season, diet and environment (Baumans, 2004). Fertilization is possible about 10–12 h after ovulation and gestation lasts for 19–21 days. Parturition usually takes place during the night, and is followed by postpartum oestrus with ovulation at 12–

18 h after giving birth (Berry, 1970), making it possible for female mice to be pregnant and raise a litter simultaneously.

1.4 Maternal behaviour

In altricial species such as the mouse, maternal behaviour is crucial for the survival of offspring. Since mouse pups have poor thermoregulatory abilities up to 2–3 weeks of age, the construction of a nest before parturition is important for successful rearing of young (Lynch & Possidente Jr, 1978).

Maternal nest building in mice starts already around day 4 after mating and the mouse thus differ from other altricial species such as rats and rabbits in that the maternal nest is prepared so early in gestation (Lisk, 1971; Lisk et al., 1969).

The nest built by the pregnant female (and sometimes also the male) before parturition differs in shape and structure from sleeping nests and is often

(13)

reported that nest size continues to increase throughout gestation until 1 day prepartum, and then gradually decreases after parturition, while other authors report a peak in the amount of nest material used around days 12–14 of pregnancy (Broida & Svare, 1982). The hormones oestradiol and progesterone seem to act in synergy to facilitate nest-building behaviour (Lisk, 1971).

During the first 12 days after birth, mouse pups are fully dependent on their mother for nutrition, temperature regulation and to stimulate defecation.

During this period the female spends most of her time close to the pups, gathering them in the nest and crouching over them, and she only leave the nest for short bouts (König & Markl, 1987). During the first 5 days after birth, the behaviours spontaneous licking, changing suckling position, and nest building decrease in female mice, and external stimuli from the pups are crucial for maintaining maternal care (Ehret & Bernecker, 1986; Cohen-Salmon et al., 1985). Pups have very limited abilities to move and the mother’s ability to retrieve the pups if they fall outside the nest is important for survival: without the insulation from the nest and the mother mouse pups rapidly lose body temperature.

1.5 Pup development and behaviour

Mouse pups are born without hair (except for whiskers), are blind, deaf, have undeveloped motor skills, only weigh approximately 1 g (König & Markl, 1987; Ewer, 1968) and typically huddle to keep warmth. They seem to start hearing by the fourth or fifth day, and by day 6 the body is covered by a thin coat of hair. The eyes open between days 12 and 14 after birth (Fuchs, 1981;

Williams & Scott, 1953) and after this they start to become active outside the nest. Except for when the pups are exploring, the eyes are often kept tightly closed until days 15 or 16. At the age of 17 days the hair coat is fully developed and the pups start to eat solid food and the weaning process gradually begins (König & Markl, 1987; Williams & Scott, 1953).

Although they are born with non-functional auditory systems (Porter, 1983), newborn rodents of several species use vocalizations (Elwood &

McCauley, 1983). During the first 2–3 weeks postpartum, pups emit a variety of ultrasonic vocalizations when they are isolated from the mother (Branchi et al., 1998). Wriggling calls increase between birth and day 5 postpartum (Ehret

& Bernecker, 1986), functioning to maintain maternal behaviour at a high level. Pups have been found to emit wriggling calls regularly during suckling, and always in association with pup movements and ultrasonic sounds can be categorised according to the response triggered in the mother (Ehret &

Bernecker, 1986). Factors in the environment can also influence the emission

(14)

of calls, such as isolation, low temperature and tactile stimulation (Branchi et al., 1998).

1.6 Postnatal mortality

Pup mortality is a considerable problem in many facilities breeding mice. Both individual pups and entire litter can be lost, and they are usually lost the first days after birth (Brown et al., 1999). Losing a small proportion of a litter or losing the whole litter is radically different in reproductive terms for the female. If only single pups are lost she will sustain lactation until the litter is weaned with more resources available for remaining pups than if no pups were lost. If the entire litter is lost she will be able to dedicate all resources to a new pregnancy avoiding competition for resources between pregnancy and lactation which may reduce litter size (McCarthy, 1965). It is difficult to get a good picture of pup mortality in research facilities. Few papers exist where mortality is systematically studied in healthy animals. Studies reporting mortality rates use different strains held under different social and physical housing conditions, and the timing and methods used to determine mortality vary. The reported mortality rates thus vary greatly between publications; from nearly none to 50% in experimental studies (Cooper et al., 2007; Whitaker et al., 2007; Inglis et al., 2004; Reeb-Whitaker et al., 2001), compared to 13%

reported for the same strain (C57BL/6) from a commercial breeder (Mouse Phenome Database, Accessed 2011).

The phenomenon of young dying shortly after birth is not unique for laboratory mice. The phenomenon is relatively well studied in farm animals where the major causes of death are similar across species: hypothermia, underfeeding, inappropriate maternal behaviour, infections and injuries (Mellor

& Stafford, 2004). High numbers of young dying has also been reported in the farmed mink with hypothermia (Malmkvist & Palme, 2008) as well as birth problems and prolonged parturition (Malmkvist et al., 2007) described to be associated with early kit mortality.

1.6.1 Infanticide and cannibalism

Pup mortality in mice is often described in terms of cannibalism or even infanticide, suggesting that the female actively kills her offspring. However, in most instances of mortality, the methodology used does not allow scientists to determine if pups were actively killed or injured or whatever other causes they

(15)

cannibalism) (McFarland, 2006). Infanticide is defined as the killing of young by conspecifics (McFarland, 2006; McCarthy & vom Saal, 1985).

The phenomenon of infanticide in terms of killing non-related offspring is well known and part of the reproductive strategy of several species. In lions for example, a new male will sometimes kill the cubs present if he takes over the pride (Krebs & Davies, 1993). The explanation for this behaviour is that losing her young will make the female receptive to mating again faster, and this gives the new male an opportunity to mate with her. Mice have been used as model organisms in laboratory studies addressing infanticide from a behavioural ecology perspective (Elwood, 1991). In these studies, males or females with differing sexual experience are exposed to related or unrelated pups and the pups are generally placed in the home cage of the test animal and behaviours measured in so called screening tests (Perrigo et al., 1993; McCarthy & vom Saal, 1986; Gandelman, 1973c; Gandelman, 1973a). Infanticidal tendencies have been reported to differ both within and between inbred laboratory strains (Perrigo et al., 1993) and wild-type mice have been reported to be more likely to exhibit infanticide (Jakubowski & Terkel, 1982; McCarthy, 1965), however screening tests are experimental set ups and not measures taken under normal husbandry conditions.

Female mice have been found to reduce litter size when food is restricted (Elwood, 1991; König, 1989), and Poley (1974) suggested that stress causes females to neglect, kill or eat their young. It should be noted, however, that under normal breeding conditions what is usually found is pups that are partly eaten, or a reduced number of pups, but no evidence of active killing.

In several papers authors refer to cannibalism as the cause of death, e.g.

“these losses were attributed to cannibalism” (Seamer & Chesterman, 1967),

“the majority of deaths (…) due primarily to cannibalism” (Morse et al., 1974),

“some of the mice displayed cannibalism toward their newborn pups” (Kang et al., 2004), “cannibalism of newborn mice by their consomic mothers was more frequent than in parental strains” (Gregorova et al., 2008), even in the absence of any information on how cannibalism was defined or observed. Others refer to the loss of pups as infanticide (Shieh et al., 2008; Stahl & Kaneda, 1999) even though there is nothing in their description of how animals were inspected that suggests they could confidently conclude that any active killing took place.

In contrast, Macbeth et al. (2010) described that pups were whole when found dead in the cage and concluded that females did not appear to attack their young; instead the underlying causes of pup deaths remained unknown.

(16)

1.6.2 Genetically modified mice

Today, thousands of different mouse models are used to study the biological functioning of mammalian genes, in 2007 Collins et al. reported that 9000 knockout models had been generated. In cases where gene mutations lead to neonatal death, pup deaths are not always a direct consequence of the primary defect, but often caused by physiological problems that arise as secondary effects (Turgeon & Meloche, 2009). Examples of morphological defects leading to neonatal death include shortened jaws and limbs, absence of limbs, lungs, eyes and nose, skeletal defects, craniofacial defects (leading to abnormal suckling), inability to open the jaws (leading to inability to vocalize resulting in rejection by the mother) (reviewed in Turgeon & Meloche, 2009). Poor maternal behaviour has been found in several models and some even show complete inability to rear offspring. Brown et al. (1996) found the fosB mutant mouse females to neglect their young, and the pups were found scattered around in the cage; the mutant mouse staggerer failed in removing the amniotic membrane, leading to pups dying from being choked, and pups that survived died from cold or hunger (Guastavino, 1984). Gαq/11-deficient females delivered pups normally but did not build nests, gather pups or crouch over them and the pups died scattered in the cage within 48 hours after birth (Wettschureck et al., 2004). Also the mutant hubb/hubb (Alston-Mills et al., 1999) and the Mecp2-deficient mouse model (Jugloff et al., 2006) are reported being difficult to breed.

1.7 Ethical concerns

In 1876, the first legislation concerning animal experimentation was set in the United Kingdom and for many years this was the only country protecting animals used for scientific purposes by legislation. The first European-wide legislation was established in 1986 with Directive 86/609/EEC. Today, animals used in countries in the European Union are protected under the Directive 2010/63/EU. According to this directive, projects where animals are part of the study must be authorised by the competent authority before the experiment can start. Prior to being approved, projects must be evaluated in an ethical review process, taking into account the ethical considerations of using animals.

Applied animal research ethics is guided by the principles of the 3Rs (Replacement, Reduction and Refinement), established more than 50 years ago by Russell and Burch (van Zupthen, 2001). The aims with the 3Rs are to use

(17)

aim is to produce genetically modified mice or the breeding itself is part of the experiment; breeding of animals is not evaluated since this is not included in the experimental set up. However, since they are laboratory animals, the principles of the 3Rs are still applied.

The work presented in this thesis focus on Reduction and Refinement. With more knowledge about what causes pup mortality in laboratory mice, the number of breeding animals needed to supply experimental animals can potentially be decreased (Reduction). Increased knowledge can also minimise suffering in both female and pups and by investigating the effect of housing environment the overall welfare might also be improved (Refinement).

(18)

(19)

2 Aims of the thesis

The overall aim with this thesis was to investigate the causes of pup mortality in laboratory mice, with focus on maternal behaviour and the effect of housing environment. Specifically the aims were to investigate:

 If litter loss is higher in primiparous than in multiparous females and if it is more likely for a female that lost a litter to lose another litter (paper I)

 If female mice actively kill their pups (paper II)

 If there are any behavioural differences between females that lose the litter shortly after birth and females that successfully wean their litters (paper III)

 How the conditions for nest building influence maternal nest building and pup survival (paper IV)

(20)

(21)

3 Materials and methods

This is an overview of the materials and methods used in the four studies included in this thesis. For full descriptions, see paper I-IV. In the first study (paper I), data from a breeding colony of laboratory mice kept at the Justus- Liebig-University of Giessen, Germany were used. The other three studies were experimental. Study 2A and 2B were conducted at the Institute for Molecular and Cell Biology, Porto, Portugal (paper II and III). These studies were carried out between June and September, 2005 (study 2A) and August 2006 and March, 2007 (study 2B). Study 3 was conducted between August 2010 and May 2011 at a large research facility in Sweden (paper IV).

3.1 Animals and housing (study 1-3)

In all studies, mice of the inbred strain C57BL/6 were used since this is the most widely used strain in research and also often used as a background strain when genetically modifying mice. In study 1 breeding data from the inbred strain BALB/c were also utilised, which is another commonly used strain. In study 2B the knockouts Hfe^-/- mice and ß2m^-/- were included. These were part of the study since the researchers using them reported problems with reproduction and were interested in investigating the effect of housing environment on their reproductive success. Females in all studies were separated from the male before giving birth and housed singly until the litter was weaned.

Animals in study 1 were housed in Makrolon III cages provided with nesting material (housing treatment S). Animals in study 2A were housed in Makrolon II cages without nesting material (B) or in Makrolon III cages with nesting material and furnishment (F; Figure 1). Animals in study 2B were housed in Makrolon II cages with a small amount of nesting material (S) or with twice the amount of nesting material and furnishment (F; Figure 1). In

(22)

study 3 animals were housed in individually ventilated cages (IVC) type 1291H in one of four treatments: small (S) or large (L) amount of nesting material, and nest structure (S) present or absent (Figure 2). For an overview of strains and housing treatments used in the different studies, see Table 1. Room conditions for all studies were standardised with temperature maintained at 19- 23 °C, relative humidity at 40-70% and a 12:12 h light/dark cycle. Animals were given ad lib access to food in the food hopper, and autoclaved water.

Animals in study 3 were also provided with food on the cage floor.

Figure 1. Housing treatments used in study 2. On top study 2A: left barren, without nesting material (B) and right furnished larger cage, half a tissue paper and furnishment (F). At the bottom study 2B: left standard, 0.5 nestlet (S) and right furnished, 1 nestlet and furnishment (F).

(Photo: Elin Weber)

(23)

Figure 2. Housing treatments used in study 3. Top left 0.5 nestlet (S), top right 3 nestlets (L).

Bottom left 0.5 nestlet and nest structure (SS), bottom right 3 nestlets and nest structure (LS).

Food provided on the cage floor. (Photo: Elin Weber)

(24)

Table 1. Overview of strains and housing treatments (cage type, bedding, nesting material and furnishment) included in studies 1-3.

Study Number of

females Strain Cage type

(size LxWxH) Bedding and

nesting material Furnishment

Housing treatment 1 111 C57BL/6 Makrolon III

(410×265×

175 mm)

Fir tree,

tissue paper None S

61 BALB/c

2A 10 C57BL/6 Makrolon II

(265×205×

140 mm)

Corncob, no nesting material

None B

10 Makrolon III

(410×265×

175 mm)

Corncob, 1 dl soft bedding, half a tissue paper

Chew block, red PVC nest box, modified cardboard nest box

F

2B 10 C57BL/6 Makrolon II

(265×205×

140 mm)

Corncob,

0.5 nestlet None S

11 Hfe^-/-

10 ß2m^-/-

10 C57BL/6 Corncob,

1 nestlet Chew block, transparent tinted mouse tunnel hanging from the grid, modified cardboard tube nest box

F

11 Hfe^-/-

10 ß2m^-/-

3 14 C57BL/6 IVC 1291H

(425×266×

185 mm)

Aspen,

0.5 nestlet None S

14 Aspen,

3 nestlets None L

14 Aspen,

0.5 nestlet Structure SS

15 Aspen,

3 nestlets Structure LS

15 Aspen,

0.5 nestlet None C^a

(control)

aControl group left undisturbed except for day 2 after birth.

(25)

3.2 Data collection

3.2.1 Pup survival (study 1-3)

For study 1, data from existing breeding records were obtained where the numbers of weaned or dead pups were indicated; losses of single pups were not recorded. There is a widespread practice of leaving periparturient females undisturbed around the time for parturition to avoid females killing their young. In study 2, females were therefore left undisturbed and the number of pups in each cage was counted at first cage cleaning after birth (study 2A day 10, study 2B day 4). In study 1 and 2, survival was thus only measured at litter level and litter loss was defined as all pups in a litter dying before weaning at day 21 after birth. However, in several studies using early handling protocols, mouse pups are handled without this procedure leading to pup loss. In study 3, the cages were therefore inspected daily and the mother and pups were also handled from day of birth. Pups were counted on day of birth (day 0) and days 1, 2, 3 and 23 after birth and the loss of both individual pups and entire litters was included in this study. In study 3 dead pups were photographed and it was noted if the pups were partly eaten, injured or intact.

3.2.2 Video recordings (study 2)

The aim of study 2 was to observe females in detail around the time of parturition. Mice are easily disturbed, they are mainly nocturnal and the exact time of parturition is very difficult to determine by only visually inspecting the cages. The females were therefore video recorded in their home cages from approximately 3 days before until 4 days after parturition. Four cages were recorded simultaneously using cameras (Ikegami ICD-47E, B/W CCD, Japan) connected to a time lapse recorder (Panasonic AG-TL750E, Thailand). The recordings were rotated by means of a camera switcher (Sanyo VQC 809-P, Japan) at 30 s intervals. In study 2B, approximately one third of the cages were recorded continuously throughout the recording period with data collected into a computer with a multi camera vigilance system (GV-800/8; GeoVision, Taipei, Taiwan).

3.2.3 Time of birth (study 2 and 3)

To determine day of birth, cages were visually inspected daily from day 18 after mating (study 2) or by removing the cage from the rack and lifting the cage lid (study 3), continuing until the litter was born. In study 2, video recordings were scanned to determine the exact time when parturition began.

After detection of pups the film was rewound and played at fast speed forward to find the female in birth position (Ewer, 1968). Time for parturition was

(26)

defined as the time when the first pup was delivered, or (if the pup was not seen) the first time when the female was seen in birth position. In cages with nesting material it can be difficult to detect the pups and see when the female starts to give birth. If neither the first pup nor birth position was possible to detect, time for parturition was estimated as the midpoint between the last time the female was seen pregnant and the first time a pup was seen or the female was seen non-pregnant.

3.2.4 Behavioural observations (study 2)

In paper II, the aim was to investigate if females actively killed their offspring.

Video recordings from females that lost their entire litter before weaning were selected and observed in detail from birth until the entire litter was lost. It is very difficult to observe females in detail when they are housed with nesting material. Therefore, only females from study 2A that were housed without nesting material (n=5) and females from study 2B with a small amount of nesting material (n=5) were selected. The Observer XT 6.1 software (Noldus Information Technology, The Netherlands) was used for scoring behaviours.

Both scan sampling and continuous observations of certain time periods were used. A pup was defined as dead when it was lying still and never seen moving again. Behavioural observations started when the first pup was born. First, movements of pups were observed immediately after birth and 1 h postpartum to determine if the pup was born alive or stillborn. Then the female and pups were observed at certain time points, using a predefined flowchart, to detect when each individual pup died. During this scan, only the behaviours “pup still” and “pup moving” were recorded. When “pup still” was observed, the pup was tracked backwards to observe what took place before it stopped moving. To establish time of death, the sequence from when a pup was last seen moving until it was still was observed in detail for all pups dying.

In paper III, the aim was to further investigate the cause of litter loss by comparing the periparturient behaviour of females that successfully weaned their litter, and females whose entire litters were lost. If more than 12 hours of recordings were missing, the females were excluded and in total 64 females from study 2 were used (study 2A n= 17, study 2B n=47). Females were observed on video from 24 hours before until 24 hours after birth and coded by one observer using a predefined ethogram (Appendix 1, paper III). Occurrence of behaviours was observed during a 30-s period every 15 min for the entire observation period (in total 48 h) by one observer.

(27)

3.2.5 Nest quality (study 3)

In study 3, nests were scored every 3rd day from day after mating until parturition, and on days 0, 1, 2 and 3 postpartum, using the naturalistic nest score developed by Hess et al. (2008). If bedding material was gathered to the site where the nest was situated, the bedding material was included when scoring the nests (Figure 3). Nest opacity and nest coverage were used as complementary measures of nest quality. Nest opacity indicated if the mice (female or pups) were visible through the nesting material or not, and nest coverage if they were visible above the edge of the nest or not, in both cases when seen from the side. Opacity and coverage were assessed from four perpendicular angles, resulting in two 5-level ordinal variables (i.e. 0=visible from all four angles; 1=visible from three angles; 2=visible from two angles;

3=visible from one angle; 4=not visible at all; Figure 3).

Figure 3. Top: Illustrations of bedding material gathered to form a nest. Bottom: a female not visible above the nest edge (illustrated with straight line) from any angle, scored as coverage 4. In the picture the edge is missing on one side, resulting in coverage 3. Arrows indicate opacity, i.e.

how visible the mice are through the nesting material. Arrow A shows visibility from one angle and arrow B no visibility from the other three angles, resulting in opacity 3. (Photo: Elin Weber)

(28)

3.2.6 Weights and health (study 3)

In study 3, human handling according to standard laboratory routines was applied daily on days 0 to 3 to achieve a relatively high level of disturbance.

During the disturbance the cage was removed from the rack and placed on a LAF-bench in an adjacent room. The female was weighed and then placed in a smaller cage while pups were individually weighed. During this separation the pups were also counted, marked with permanent ink (Promarker, Letraset, UK), checked for milk spots, and the cages inspected for dead pups which, when found, were removed from the cage. Female and litter were weighed again at weaning on day 23. To estimate the effect of disturbance a fifth group of animals, with a small amount of nesting material and no access to nest structure, was left undisturbed except for day 2 when the pups were counted and pups and female were weighed (control).

3.2.7 Pilot study (study 3)

A pilot study was conducted to examine the possibility of using thermal imaging to measure heat loss in mouse pups. During the daily disturbance from day 0-3, cages where placed under a thermal camera (ThermaCAM S60, FLIR, US) and heat radiation was measured during 1 minute.

3.3 Data analysis

3.3.1 Study 1

In total, 344 litter observations from 111 parental couples from 12 breeding groups of C57BL/6 and 146 litters bred by 61 parental couples from seven breeding groups of BALB/c were included in the analysis. All females in a breeding group originated from the same breeding pair and each parental couple contributed with between 1 and 8 litters (median 3). Litter loss referred to whole litter being lost, and was coded as a binary outcome (0=litter not lost;

1=litter lost). The risk of litter loss was modelled using a generalised linear model in the GENMOD procedure of SAS (version 9; SAS Institute Inc., Cary, NC, USA) and the clustering of litters from the same parental couple was accounted for. Fixed-effect predictors were constructed expressing strain (C57BL/6, BALB/c) and parity (primiparous, multiparous), and whether or not there was a previous record of litter loss in the same parental couple (no, yes).

The final model contained strain and parity effects, and the interaction between strain and parity. Model-based marginal means were calculated to estimate the

(29)

3.3.2 Study 2

In paper II the course of events from birth of a litter until all pups were dead were observed and described in detail. Besides describing the interactions between female and pups, other events (i.e. aberrant behaviours, problematic parturition) that might be relevant for pup survival were also described.

In paper III data were arranged with one observation per 15-min observation and the observation period was divided into sub-periods (48 1-h, 16 3-h, 8 6-h, and 2 24-h periods). Data were averaged for each female, sub- period and for the entire observation period, by calculating female-specific relative frequencies of all behaviours. Some behaviours occurred in low frequency and where therefore aggregated into behaviour categories (Table 2).

Table 2. Overview of behaviours from the ethogram aggregated into behaviour categories.

Behaviour category Behaviours included

Parturition-related Giving birth, Labour, Dystocia

Abnormal Removing pup, Resting alone, Ignoring active

pup, Ignoring still pup, Resting outside nest, Hunched posture, Digging, Tail chasing, Bar gnawing, Removing pup, Other abnormal Self-oriented Self-grooming, Resting alone, Ignoring active

pup, Ignoring still pup, Resting outside nest, Hunched posture, Digging, Tail chasing, Bar gnawing

Nest building Nest building, Move nest

Active maternal behaviour Active with pup, Retrieve still pup, Retrieve active pup, Carrying pup, Moving pup, Active in nest

Passive maternal behaviour Nursing, Still in nest

To examine the association between survival and predictors representing study (2A or 2B), cage design (furnished or not), and mouse strain (C57BL/6, Hfe^-/- or β2m^-/-), simple logistic regression models of litter survival (no or yes) were constructed at the female level using the Stata Logit command (StataCorp SLP, College Station, Texas, USA), containing one predictor variable at a time. To investigate the association of survival with different behaviours, ten hypotheses were formulated based on the aggregated behaviour categories as well as the behaviours nest building, being outside nest and ignoring still or active pup. A simple logistic regression model of survival was constructed for each

(30)

hypothesis, containing only one of the aggregated behaviour variables. In a second analytical step, behaviour variables that were found to be significantly associated with survival at p≤0.05 were used to construct a multivariable logistic regression model of survival.

3.3.3 Study 3

Nest quality was analysed by modelling the three outcome traits (nest score, opacity, coverage) separately. Nest score was normalised by calculating the natural logarithm of the reversed original score, i.e. ln(6 – score). It was analysed by mixed-effects linear modelling using the Stata Mixed command.

There were few observations with low scores for the traits opacity and coverage, the ordinal dependent variables were therefore in both cases obtained by collapsing the two lowest levels, thus creating two variables with four levels (1, 2, 3 and 4), analysed by ordinal logistic modelling. Month, day and hour of the day were re-coded as categorical independent variables, each with four approximately equally-sized categories. Pairwise correlations among the three dependent variables were checked. Each trait was modelled to estimate the effects of nesting material and access to nest structure from 21 days before to 4 days after parturition (comparing treatments S, L, SS and LS). In all three models, day category was included, as well as potential confounders.

3.4 Ethical approval (study 2-3)

Study 2 was carried out under a project license (ref. 003758) issued by the Direcção Geral de Veterinária, the competent authority for animal protection in Portugal. Study 3 was approved by the Swedish Regional Ethics Committee for animal experiments.

(31)

4 Summary of results

This section summarises the main results of study 1-3, more details can be found in paper I-IV. An effect of strain but no effect of parity on litter mortality was found. No evidence that females actively killed their pups was found. Litter loss was mainly associated with females showing less nest- building behaviour before parturition and more time outside the nest. Females given a large amount of nesting material built dome shaped nests of higher quality. The total survival of litters in all studies is shown in Figure 4.

Figure 4. Overview of survival for all studies (n=number of litters born). In study 1 and 2B, litters from 1^st and up to 8^th and 7^th parity, respectively, are shown. For study 2A and study 3, only 1^st parity litters were included. Red=litters dead before weaning, green=litters survived until weaning at around 3 weeks. C57=strain C57BL/6 and BALB=strain BALB/c. B=no nesting material provided, S=small amount of nesting material provided, F=nesting material and furnishment, SS=

small amount of nesting material and structure, L=nesting material, LS=nesting material and structure (details in Table 1).

100 2030 4050 6070 8090 100

S (n=344) S (n=146) F (n=9) B (n=10) F (n=32) S (n=38) F (n=36) S (n=36) F (n=26) S (n=31) S (n=14) SS (n=12) L (n=14) LS (n=14)

C57 BALB C57 C57 Hfe-/- ß2m-/- C57

Study 1 Study 2A Study 2B Study 3

Percent of litters

(32)

4.1 Study 1 (paper I)

An overall high mortality was found in the breeding record for both strains, with a total mortality rate (calculated as loss of entire litters) of 32% for C57BL/6 and 20% for BALB/c (Table 3). A statistically significant effect of strain was found in the first parity, in that primiparous C57BL/6 females were more likely to lose their litters than primiparous BALB/c females (p=0.0028).

No other effects of parity or loss of earlier litters on litter loss could be found.

Table 3. Distribution by strain and parity, and litter loss in 490 laboratory mouse litters of C57BL/6 or BALB/c strains in study 1.

Strain Parity No. of litters No. of litters lost (%) C57BL/6

BALB/c

1 2 3 4 5 6 7 8 1 2 3 4 5 6

111 90 62 36 25 14 5 1 61 45 26 8 4 2

39 (35) 27 (30) 16 (26) 10 (28) 11 (44) 4 (29) 3 (60) 0 (0) 8 (13) 12 (27) 6 (23) 2 (25) 1 (25) 0 (0)

4.2 Study 2 (paper II and III)

4.2.1 Paper II

In paper II three females had entire litters in which pups were never seen moving. Another three females had 1-2 pups that were never seen moving.

This indicates that some pups were most likely dead at birth. While scoring interactions between mother and pups several observations of the females were made that indicated problems of giving birth. In one female, the first pup was stuck for 1 h in the birth canal during parturition. This pup was never seen moving and the female did not interact with the pup after it came loose. The female was lying in a hunched posture outside the nest for several hours before parturition (Figure 5) and was also outside the nest when the parturition started and during the following 30 min. Another female was lying outside the nest in

(33)

nest and pups to a new location in the cage about 1.5 day after parturition, but moved it back to the original site 3 h later.

Figure 5. Images from video recording showing one female that was lying in a hunched posture for several hours before giving birth (left), and later had problems during parturition. The first pup was born 10 hours after this picture and stuck in the birth canal for 1 h after parturition started.

Another female (right) was lying outside the nest for several hours while her live pups (indicated with arrows) were scattered in the nest material. (Photo: Elin Weber)

Detailed observations of interactions between mother and pups were possible to carry out for at least one pup per female in the seven females with live born pups. Females were interacting with both still and moving pups, and were observed performing maternal behaviours (e.g. licking and retrieving) towards dead pups (Figure 6). Females were also observed eating dead offspring (sometimes while still having live pups in the nest), but on no occasion was a female observed manipulating a moving pup that stopped moving directly after the manipulation without moving again. In most cases the pups displayed successively smaller movements until their activity was very difficult to detect and rarely seen, and the pups were finally lying still not moving anymore.

Females were not observed eating pups immediately after they had stopped moving. In most cases the pups were lying still for several hours before the female started eating them (Figure 6).

(34)

Figure 6. Images from video recordings showing a female with dead pups. Only three very small

(35)

4.2.2 Paper III

Behaviour was analysed in 64 females, of these 49 successfully weaned a litter (study A n=12; study B n=37) and 15 had litters that died (study A n=5; study B n=10). Several significant associations between behaviours and survival were found (Table 4). Survival of the litter was associated with the females showing more nest-building behaviour during the last 24 h before parturition (p=0.004) and being less outside nest between 24 h before and 24 h after parturition (p=0.001). Increased litter survival was also associated with females performing more passive maternal behaviours (p=0.006) and ignoring still pups less 24 h after parturition (p=0.035). Females that lost their litters performed more parturition-related behaviours during the last 6 h before giving birth (p=0.020). A final multivariable logistic model of survival contained the behaviours “nest building” before parturition and “outside nest”, and these together accounted for 33% of the variation in survival. Predictive marginal means with 95% confidence intervals are shown in Figure 7.

Figure 7. Predicted probability of litter survival as a function of proportion of observations with nest building and being outside nest, the two behaviours most strongly associated with low litter survival. Shaded area represents 95% confidence interval.

.4.6.81Probability of survival

0 .02 .04 .06 .08 .1

Prop. of observations Nest building

0.2.4.6.81Probability of survival

.1 .15 .2 .25

Prop. of observations Being outside nest

(36)

Table 4. Summary of simple logistic regression models of the five behaviours for which significant linear associations were found.

Behaviour Coef. Std. Err. OR¹ P>|z|

Nest building 76.68 26.50472 2.2 0.004

Outside nest -26.63 7.976263 0.77 0.001

Parturition-related -11.93 5.129796 0.89 0.020

Passive maternal behaviour 9.60 3.515016 1.1 0.006

Ignore still pup -3.56 1.687502 0.96 0.035

1OR=change in odds of survival per percent unit increase in frequency of behaviour.

4.3 Study 3 (paper IV)

Of the 66 females that conceived and gave birth to a litter, 12% lost their entire litter before day 2 after birth (treatments S and SS, 15.4%; treatments L and LS, 10.7%) and another 17% lost part of their litter (1-4 pups). The majority of pups died on day 0 or 1.

Females given a large amount of nesting material (L and LS treatments) built larger nests than females with a small amount of nesting material (S and SS treatments) (Figure 8). The nests were of higher quality with regard to the naturalistic nest score, nest opacity and nest coverage.

(37)

A large amount of nesting material resulted in significantly higher predicted nest scores at all day categories (p<0.001), compared to a small amount and access to nest structure increased nest score by between 0.14 and 0.20 (p=0.007) (Figure 9). The higher the nest score, the more complete is the nest, with the highest score of 5 corresponding to a completely closed dome shaped nest (Figure 8, bottom right). A large amount of nesting material decreased the odds of incomplete opacity by 98% (p<0.001) and of incomplete coverage by 99% (p<0.001) across all levels of the traits (Figure 10). Incomplete opacity indicates a nest where the female or pups are visible through the nest material and incomplete coverage indicates a nest where the female or pups are visible above the edge of the nest.

Figure 9. Predicted margins of maternal nest scores across days according to a mixed-effects regression model. Large amount of nesting material (3 nestlets) and access to nest structure (black solid line), large amount of nesting material and no nest structure (black dashed), small amount of nesting (0.5 nestlet) material and access to structure (grey solid), and small amount and no access to structure (grey dashed).

012345Nest score

0 to 4 -8 to -1

-15 to -9 -21 to -16

Day relative to parturition

(38)

Figure 10. Predictive probability of maternal incomplete nest coverage (below score 4; top) and nest opacity (below score 4; bottom) across day categories with small amount of nesting material (0.5 nestlet) and no access to a nest structure (grey dashed), small amount and access to structure (grey solid), large amount (3 nestlets) and no access to structure (black dashed), and large amount and access to structure (black solid line), according to an ordinal logistic regression model in

0.2.4.6.81p(Nest opacity <4)

0 to 4 -8 to -1

-15 to -9 -21 to -16

0.2.4.6.81p(Nest coverage <4)

0 to 4 -8 to -1

-15 to -9 -21 to -16

(39)

One female was found in a very poor condition (Figure 11) and had to be euthanized on day 1; three of her pups were found dead on day 0, one pup was alive on day 0 but found dead on day 1 and three unborn pups were found in the post mortem analysis of the female. Similar to study 2, dead pups were found in the outer edge of the nest (Figure 12). When the cages were inspected for dead offspring, pups were found remaining in the amniotic sac, partly eaten or intact, but no visible wounds were found in the intact pups (Figure 13).

Dead pups differed from the live pups in colour (they were pale or grey), temperature (they were often cold) and activity (lying totally still). The dead pups were often found in the bedding material under the nest, but sometimes dead pups were lying with the rest of the litter in the nest (Figure 14).

Figure 11. Female found in a very poor condition and was euthanized on day 1; dead pups were spread out around the nest. (Photo: Anne Larsen)

(40)

Figure 12. Female mouse in nest, two dead pups (indicated with arrows) have been pushed to the outer edge of the nest. (Photo: Elin Weber)

(41)

Figure 14. Top pictures illustrating dead pups found in the bedding material at cage inspection, bottom left picture showing dead pup found together with the live litter in the nest. Dead pups are marked with arrows in pictures with several pups or when hidden in the bedding.

(Photo: Elin Weber)

(42)

4.3.1 Pilot study

Thermal imaging was feasible on newborn mouse pups as they were less mobile than adults. Measurements could not be made through the plastic cage;

images were therefore obtained from above after the cage lid was removed. It was possible to follow thermal radiation both from the litter and from single pups outside the nest (Figure 15), and to detect differences in thermal radiation over the measured period of 1 min. However, if pups were not alive, they were not possible to detect since they had no heat radiation and did not differ from the surrounding bedding material.

Figure 15. Caption from thermal imaging illustrating a mouse female, the litter and one pup outside nest. (Photo: Elin Weber)

(43)

5 Discussion

The focus of this thesis is to increase the understanding of pup mortality in laboratory mice, on the background of the general assumptions that it is normal for mice to lose their first litter and for some mothers to kill their offspring after birth. No support for these assumptions has been found and in the following sections the different questions raised in the included studies will be addressed, followed by a general discussion on pup mortality in laboratory mice. Problems with pups dying within the first days after birth complicates planning of research; if pups are lost there is an uncertain number of animals available for research. In several breeding facilities this has led to the practice of maintaining additional breeding animals to compensate for pup mortality, which in turn leads to an increased workload and cost. Mice are small animals and several breeding females can be kept together, and keeping additional breeding animals does not require much extra space. When discussing pup mortality with breeders and reviewing the scientific literature, it becomes clear that empirical data on pup mortality are scarce. Generally, what is described as mothers that have killed their pups often turns out to be observations of partly eaten pups or disappearance of previously observed pups without any direct observations of mothers actively killing their pups. Even in the scientific publications infanticide and cannibalism are described as causes of death without data supporting these conclusions. Evidence of poor survival of first litters is also limited.

5.1 Influence of strain and parity

When comparing the inbred mouse strains C57BL/6 and BALB/c in study 1 (paper I), a high percentage of entire litters lost (32% and 20% respectively) was found in both strains. However, high pup mortality in C57BL/6 has previously been reported (Gaskill et al., 2013a, 30%; Brown et al., 1999, 36%;

Potgieter & Wilke 1997, 22.4%). The survival of first litters differed between

(44)

strains with C57BL/6 females having a higher mortality rate in their first litters compared to BALB/c. Differences between strains have been described for a wide variety of traits, including reproductive performance (Brown et al., 1999;

Potgieter & Wilke, 1997). However, in this study, a difference was only found in the first litter; there were no strain differences in overall survival across parities. An inability of primiparous female mammals to care appropriately for their offspring has been described, with maternal responsiveness reported to affect survival (Nowak et al., 2000). Although Brown et al. (1999) found higher survival in second than in the first litters in both C57BL⁄ 6J and DBA⁄

2J mice, no effect of parity in any of the strains could be found in this study.

This discrepancy in results might be explained by mortality calculated as loss of entire litters in this study, compared to loss of single pups in the study by Brown et al. (1999).

5.2 Infanticide

Under certain circumstances it can be adaptive for a female to kill her offspring, if killing of one or more offspring increase the chance of weaning the remaining litter (Elwood, 1991). In a study investigating cannibalism in the golden hamster (Mesocricetus auratus), Day and Galef (1977) found that female hamsters adjusted their litter size to a specific size the first days postpartum. They concluded that it was a reproductive strategy for the females to adjust number of young to their capacity to wean them. König (1989) further found that when under food restriction, female mice killed part of their litter.

To examine if female mice killed their pups, dams were observed in detail from time of birth until the pups died (paper II). They were observed interacting with both live and dead pups, but were never observed actively killing their young. Instead, they displayed maternal behaviours with dead pups such as retrieving them to the nest, crouching over them and licking them.

Some pups were never seen moving and were thus likely stillborn. Others were observed spread out inside or outside the nest and gradually decreasing movements were observed until the pups eventually stopped moving and remained still. After the pups stopped moving they were lying still for hours before the female began to consume them. To eat dead offspring could be considered adaptive; a dead pup constitutes energy and also, if dead pups are not removed from the nest site it will eventually lead to unhygienic and unhealthy conditions. In the confinement of the laboratory cage a female

(45)

are housed under a normal dark:light schedule (i.e. not reversed) and a female gives birth during the night, dead pups might be present in the cage for several hours before laboratory personnel removes them from the cage. During the behavioural observations it was found that dead pups could be lying intact in the cage for several hours before the female started to consume them. Also in the farmed mink (Mustela vison) maternal infanticide has been suggested to be among the main causes of perinatal mortality. However, in a detailed study on periparturient behaviour, Malmkvist et al. (2007) found no evidence of infanticidal mothers.

5.3 Influence of behaviour

Several studies have investigated the effect of different factors (e.g. strain, housing systems, nesting material) on maternal behaviour (Shoji & Kato, 2006;

Brown et al., 1999) and reproductive performance (Spangenberg et al., 2014;

Gaskill et al., 2013b; Carvalho et al., 2009; Rasmussen et al., 2009; Tsai et al., 2003; Bond et al., 2002; Eskola & Kaliste-Korhonen, 1999; Potgieter & Wilke, 1997), and differences both in terms of maternal behaviour and survival of offspring have been reported. However, paper III in this thesis is the first study in mice that compare the behaviours of females that lose their entire litters before weaning with females who successfully wean their litters. It was found that females from the two groups differed in several of the behaviours observed. Females that successfully weaned their litters performed more nest- building behaviour the day before parturition. This result is in line with previous research stressing the importance of a nest of high quality (Brown, 1953) and access to nesting material for survival of offspring (Gaskill et al., 2013b). Losing a litter was further associated with females being more outside nest both before and after parturition, as well as the female being less passive inside the nest. Mouse pups are fully dependent on their mother and on insulating properties of the nest for nutrition and maintenance of body temperature; to be born in a protected environment is thus crucial for survival.

To prepare a nest before giving birth and spending more time inside the nest decrease the risk of pups losing body temperature and increase survival. Being more passive inside the nest might facilitate for pups to find their way to the nipples and suckle for longer periods. Furthermore, a moderate amount of active maternal behaviour was found to be associated with maximum survival.

A combination of being still inside the nest and active during certain periods may thus be optimal for proper caretaking of the pups. Licking is an example of active maternal behaviour and an important component of maternal

Pup Mortality in Laboratory Mice